The Research On The Real-time Image Processing Of Weld Vision Tracking Based On OpenCV

Along with the continuing rapid development in machinery manufacturing,petrochemical, metallurgy, aviation and other fields, welding is regarded as the key processesin the industrial manufacturing sector.Because directly affecting the product quality andproduction efficiency, it has attracted more and more social attention. In a specific productionpractice, the high level of automation of welding equipment make the real time seam trackingwelding research become the main research topics of the field of welding automation.Somany domestic and foreign research institutions and scholars have devoted much energy tothis field.Basing on the analysis of the status of domestic and foreign welding seam automaticidentification technology and the communication with related experts in the field, this paperdesign a set of real time seam processing system based on OpenCV software. The system usesa CCD camera as the weld inspection sensor to capture images in real-time welding,introductcomputer processing technology and machine vision technology, use image acquisition cardtoput real-time images from CCD camera into the computer memory and do the imageprocessing operation.It will help improve the weld image processing speed.The above signalis put into PLC to control the implementing agencies to move, and ultimately achieve thepurpose of welding automation.The entire central part of the software system is the image processing part. OpenCVsoftware is free for any customer，at the same time it is an excellent image processing andmachine vision software.It has an unparalleled advantage with a good interface together withthe acquisition card and the PLC, and can achieve a good treatment effect. This papercompare characteristics,advantages and disadvantages of the simple smoothing filtering, theGauss filtering and the median filtering, and eventually adopt the algorithm of median filterfor image filtering. As the site environment and material quality is different, so in two valueof the image processing we need automatically choose the threshold.In the study of previousresearch foundation, this article choose Otsu method adaptively select the threshold of the twovalues of images. At the same time, in contrast to Sobel, Prewitt, Robert, Log, zero cross,Canny and wavelet algorithm in the edge detection,and according to the actual weldingcharacteristics,this paper select and optimize the Canny edge detector, then use the left andright scanning method to get weld line information.Through the experiment,the imageprocessing software effect is better,reaching set requirements.The weld seam tracking system Based on OpenCV and CCD in this paper has the second development ability accord to the change of the weld environment. If it can be appliedspecific automation welding control, can not only improve productivity and guarantee thequality of products, but also improve the working environment of the workers, pull theworkers out the first-line hard conditions, and achieve the liberation of productivity function.